On the ninth day of OpenAI’s holiday promotional campaign, the organization revealed that it would be rolling out the complete version of its o1 reasoning model to a select group of developers via its API. Prior to this announcement, developers were limited to the less advanced o1-preview version.
The company stated that the full o1 model will be gradually available to users classified as “Tier 5” developers. This tier includes individuals who have maintained an account for over a month and have spent a minimum of $1,000 with OpenAI. Given the extensive computational resources needed for the o1, the service comes at a premium price: $15 for approximately 750,000 words analyzed and $60 for generating the same quantity of text. This pricing is significantly higher—three to four times—compared to similar tasks conducted using GPT-4o.
At these elevated prices, OpenAI ensured that the complete model boasts enhanced capabilities compared to the preview version. The new o1 model allows for more customization; users can leverage a new parameter called “reasoning_effort” to control how long the AI contemplates a question. Additionally, features such as function calling, developer messages, and image analysis—absent from the o1-preview—are now included.
OpenAI also shared that it will integrate its GPT-4o and 4o-mini models into its Realtime API, specifically designed for applications requiring low-latency vocal interaction (like the Enhanced Voice Mode). The API is now compatible with WebRTC, the standard technology for developing vocal-AI applications in web browsers. Expect a surge of websites with interactive voice features by 2025.
“Our WebRTC integration is intended to provide seamless interactions even in fluctuating network conditions,” OpenAI highlighted in their announcement. “It ensures efficient audio encoding, streaming, noise cancellation, and congestion management.”
During the live-stream event, OpenAI has also unveiled the complete o1 model, introduced the Sora video generation model, launched a new Projects feature, and provided numerous updates to its Canvas, Search, and Enhanced Voice Mode functionalities.
With just three days remaining until the conclusion of the event, we are eager to see what innovations OpenAI will unveil next. Stay tuned!
